Brain Research 1001 (2004) 108–117 www.elsevier.com/locate/brainres

Research report Expression of the transporter ferroportin in synaptic vesicles and the blood–brain barrier

Laura Jui-chen Wua,b, A.G. Miriam Leendersc, Sharon Coopermana, Esther Meyron-Holtza, Sophia Smitha, William Landa, Robert Y.L. Tsaid, Urs V. Bergere, Zu-Hang Shengc, Tracey A. Rouaulta,*

a Biology and Branch, National Institute of Child Health and Human Development, Bldg. 18T, Room 101, NICHD, Bethesda, MD, 20892, USA b Chang Gung Memorial Hospital, Taiwan c Synaptic Function Unit, National Institute of Neurological Disorders and Stroke, Bethesda, MD, USA d Laboratory of Molecular Biology, National Institute of Neurological Disorders and Stroke, Bethesda, MD, USA e UB-In Situ, Natick, MA, USA Accepted 16 October 2003

Abstract

Iron homeostasis in the mammalian brain is an important and poorly understood subject. -bound iron enters the endothelial cells of the blood–brain barrier from the systemic circulation, and iron subsequently dissociates from transferrin to enter brain parenchyma by an unknown mechanism. In recent years, several iron transporters, including the iron importer DMT1 (Ireg1, MTP, DCT1) and the iron exporter ferroportin (SLC11A3, Ireg, MTP1) have been cloned and characterized. To better understand brain iron homeostasis, we have characterized the distribution of ferroportin, the presumed intestinal iron exporter, and have evaluated its potential role in regulation of iron homeostasis in the central nervous system. We discovered using in situ hybridization and immunohistochemistry that ferroportin is expressed in the endothelial cells of the blood–brain barrier, in neurons, oligodendrocytes, astrocytes, and the choroid plexus and ependymal cells. In addition, we discovered using techniques of immunoelectron microscopy and biochemical purification of synaptic vesicles that ferroportin is associated with synaptic vesicles. In the blood–brain barrier, it is likely that ferroportin serves as a molecular transporter of iron on the abluminal membrane of polarized endothelial cells. The role of ferroportin in synaptic vesicles is unknown, but its presence at that site may prove to be of great importance in neuronal iron toxicity. The widespread representation of ferroportin at sites such as the blood–brain barrier and synaptic vesicles raises the possibility that trafficking of elemental iron may be instrumental in the distribution of iron in the central nervous system. Published by Elsevier B.V.

Theme: Cellular and molecular biology Topic: Blood–brain barrier

Keywords: Iron transport; Synaptic vesicles; Endothelial transport; Neurodegeneration

1. Introduction (CNS). It is involved in myelin formation [7], as well as in the production of several neurotransmitters such as dopamine Iron is an essential element for normal cellular functions. [19], norepinephrine and serotonin [7], and generation of Many in the electron transport chain, including GABAergic activity [16]. In addition, is impli- cytochromes a, b, and c, and cytochrome oxidase, use iron cated as a cause of neuronal death [3]. Abnormally high levels as a cofactor for adenosine triphosphate (ATP) synthesis [11]. of iron in the brain have been demonstrated in a number of Iron also plays specific roles in the central nervous system neurodegenerative disorders, including NBIA 1 disease (for- merly known as Hallervorden-Spatz syndrome) [13], Parkin- * Corresponding author. Tel.: +1-301-496-7060; fax: +1-301-402- son’s disease [3], autosomal recessive juvenile parkinsonism 0078. [25], multiple system atrophy [8], Alzheimer’s disease [24] E-mail address: [email protected] (T.A. Rouault). and Friedreich’s ataxia [28].

0006-8993/$ - see front matter. Published by Elsevier B.V. doi:10.1016/j.brainres.2003.10.066 L.J. Wu et al. / Brain Research 1001 (2004) 108–117 109

Despite the importance of iron for normal brain function, the mouse ferroportin 1 peptide sequence, aa 165–181 iron homeostasis in the CNS is still not well understood (re- (ENRSRLADMNATIRRID, MAP23) or aa 240–258 viewed in Ref. [20]). Iron preferentially accumulates in the (KVEESELKQLTSPKDTEPK, MAP 24) or aa 432–447 striatum of aging animals, but the causes and significance of (EMHMSNMSNVHEMSTK, MAP 25). Anti-MAP anti- iron accumulation remain unclear [17]. Although much bodies were purified using affinity chromatography col- remains to be learned about the mechanisms responsible for umns made by coupling the MAP antigen to cyanogen- iron uptake, transport, and export across the blood–brain- bromide-activated Sepharose CL-4B. barrier (BBB) and within brain parenchyma, the recent identi- fication and characterization of several membrane iron trans- 2.2. Immunohistochemical studies porter may provide new insights into these processes. Mice of various genotypes and dietary treatments were Prior studies of the iron importer DMT1 and iron exporter intracardially perfused, post-fixed in formalin, embedded in ferroportin 1 have demonstrated their importance in transport paraffin and sectioned as described previously [15]. Non- of iron across the duodenal mucosa and in iron homeostasis specific binding was blocked with 10% goat serum and (reviewed in Ref. [2]). In the duodenal epithelium, DMT1 0.3% Triton  100 in 1  PBS for one hour at RT. localizes to the apical membrane [6] where it mediates uptake Subsequently, these sections were incubated with antiMAP of iron from the intestinal lumen, whereas ferroportin is antibodies to ferroportin (5 Ag/ml ) in blocking buffer over present mainly in the basolateral membrane [9], where it is night at 4 jC. The slides were washed with 1X PBS and presumed to export elemental iron to the circulation. The incubated with the secondary antibody biotinylated antirab- ability of these two proteins to import and export iron across bit IgG (1: 1000, Vector Laboratories) followed by Avidin the plasma membrane has also been demonstrated in Xen- Biotin Complex labeled with horseradish peroxidase (ABC opus oocytes [12] and cultured mammalian cells [1]. DMT1 Elite kit, Vector Laboratories ). HRP was visualized using transcripts have been detected by in situ hybridization in the the substrate diaminobenzidine (DAB, Vector Laboratories) choroid plexus and in neurons throughout the CNS. Ferro- with incubation for 5 min at RT. Peptide competitions were portin has previously been detected in pyramidal neurons and used as negative control. Doublelabeling immunofluores- cerebellar Purkinje cells, but not in the choroid plexus or cence studies were carried out by incubating the brain blood–brain barrier [5]. sections with primary antibody to ferroportin and primary In mammalian cells, there are two cytosolic proteins, iron- antibodies to neuronal nuclei (Neu N, mouse monoclonal regulatory proteins 1 and 2 (IRP1 and IRP2), that regulate antibody 1:100, Chemicon, CA), to astrocytes (GFAP, expression of iron metabolism proteins, such as and mouse monoclonal antibody 1:100) and to oligodendrocytes , by binding to a RNA stem loops known (CNPase, mouse monoclonal antibody 1:100) respectively as iron-responsive elements (IREs) within transcripts of those followed by incubation with fluorconjugated secondary proteins (reviewed in [21,22]. Clearly IRP2 has an important antibodies against rabbit or mouse IgG. role in CNS iron metabolism, as evidenced by the fact that IRP2 À / À mice iron accumulate iron in characteristic 2.3. In situ analysis regions of the brain that undergo neurodegeneration [15]. Ferroportin transcripts contain an IRE within the 5VUTR, but Non-isotopic in situ hybridization was performed using the regulation of ferroportin through IRPs is incompletely digoxigenin (DIG)-labeled cRNA probes and alkaline phos- understood [1]. Evaluation of ferroportin regulation and phatase (AP) detection as described [4]. Briefly, cryostat expression in the IRP2 À / À animal model therefore has sections of fresh frozen brain were cut at 10-Am thickness, the potential to provide insight into the mechanisms by which fixed in 4% paraformaldehyde and acetylated. Hybridization neurons regulate iron homeostasis. was performed in slide mailers by total immersion in In this study, we analyzed the distribution of ferroportin in hybridization buffer that contained 50% formamide, 5  the CNS of wild type and IRP2 À / À mice, and compared its SSC, 2% blocking reagent (Roche Applied Science, Indian- expression levels in iron overload or depletion paradigms. We apolis, IN), 0.02% SDS, 0.1% sarcosine, and approximately also discovered that ferroportin is expressed in synaptic 100 ng/ml of cRNA probe. Sections were hybridized at 68 vesicles, which may suggest a role of ferroportin in the jC over 72 h with the full-length mouse ferroportin probe regulation of neurotransmitter synthesis and/or synaptic (3kb) that had been alkali-hydrolyzed to around 500 bases function by iron. in length. Washing steps included incubations in 2  SSC and 0.2  SSC at 68 jC. Sections were incubated at room 2. Material and methods temperature in 1% blocking reagent in maleic acid buffer, then in AP-conjugated anti-digoxigenin Fab fragments 2.1. Antibodies (1:5000 dilution, Roche), and developed overnight with BCIP/NBT substrate (Kierkegard and Perry Laboratories, Antibodies were raised against multiple antigenetic pep- Gaithersburg, MD). Sections were rinsed several times in tides (MAPs) [27] as previously described [15], including 100 mM Tris, 150 mM NaCl, 20 mM EDTA pH 9.5, and 110 L.J. Wu et al. / Brain Research 1001 (2004) 108–117 coverslipped with glycerol gelatin (Sigma). Control sections 2.7. Synaptosomal preparations were incubated in an identical concentration of the sense probe transcript. Some slides were double-stained after the Synaptosomes from rat brain were prepared from 3 in situ hybridization procedure with a rabbit anti-GFAP week old rats maintained on a normal rodent chow diet antibody (DAKO) to examine expression in astrocytes. by differential and discontinuous Percoll gradient centrifu- The anti-GFAP antibodies were visualized using a CY3- gation [10,23], and subcellular fractions were isolated on labeled secondary antibody. sucrose gradients as previously described [14]. Eight rat brains from three-week old animals maintained on normal 2.4. Northern analysis mouse chow were processed for synaptosome preps. Brief- ly, after Percoll-sucrose gradient centrifugation, the synap- Total RNA was isolated from various mouse brain tissues tosomes were washed once in medium M (0.32 M sucrose, using Trizol reagent (Life Technologies ). Five micrograms 1mMK2HPO4, 0.1 mM EDTA [pH 7.5]), resuspended total RNA were separated in 4% formaldehyde/1.2% aga- and homogenized in medium L (1 mM K2HPO4, 0.1 mM rose gels and transferred to a Nylon membrane. 32P-labelled EDTA [pH 8.0]), and incubated at 4 jC for 1 h. The ferroportin probe was synthesized using random primer kit. Hybridization and washing conditions were as described previously, and transferrin receptor mRNA was detected as previously described [15].

2.5. Western blot

Western blot experiments were carried out as previously described [15]. In short, proteins were extracted in lysis buffer (0.125 M Tris, pH 6.8, 70 mM dodecyl sulfate, 2% Glycerol, 10 Ag/ml leupeptin, 10 Ag/ml apro- tinin, 10 Ag/ml AEBSF ). Samples were sonicated and assayed for concentration by BCA protein assay. 50 Ag of protein was mixed with an equal volume of 4 Â sample-loading buffer (0.5 M Tris, pH 6.8, 10% glycerol, 2% SDS, 0.7 M h-mercaptoethanol and 0.05% bromophe- nol blue), and fractionated in an 8% SDS acrylamide gel and transferred to a nitrocellulose membrane. The mem- brane was blocked with 5% non-fat milk in PBS and incubated with an antibody directed against ferroportin or DMT1. The membrane was stained with a horse-radish peroxidase-conjugated goat anti-rabbit IgG antibody (1:5000 dilution ) and developed using enhanced chemilu- minescence (ECL kit, Pierce).

2.6. Pre-embedding electron microscopic immunocytochemistry

Brain sections or cultured cells were fixed with 4% paraformaldehyde in 0.1 M buffer at pH 7.4 for 45 min, washed with buffer, permeabilized with 0.1% saponin and blocked with 5% normal goat serum in 1 Â PBS for 1 h. They were then incubated with the primary antibody for 1 h, washed, incubated with the Fig. 1. (A) Antibodies to ferroportin peptides detect a single 62kD band. secondary antibody conjugated to gold (Nanogold, Nanop- Predicted topology of ferroportin showing nine transmembrane domains robes, Yaphank, NY) for 1 h, washed, and fixed with 2% and the position of peptides used to make antibodies as described in glutaraldehyde in 1 Â PBS for 30 min. Samples were then Materials and methods. Antibodies were raised against 165– washed and silver enhanced (HQ silver enhancement kit, 181 (MAP 23), amino acids 240–258 (MAP 24), and against amino acids Nanoprobes), treated with 0.2% osmium tetroxide in buffer 432– 447 (MAP25). (B) Western blot analysis using three different antibodies shows a major band around 62 kD. Fifty micrograms protein for 30 min, 0.5% uranyl acetate for 30–60 min or from brain lysate were used per lane. Lane 1: MAP 23, lane 2: MAP 24, overnight, washed, dehydrated in ethanol and finally lane 4: MAP 25, lane 4: negative control with excess MAP 24 peptide embedded in epoxy resins and sectioned conventionally. competition. ..W ta./BanRsac 01(04 108–117 (2004) 1001 Research Brain / al. et Wu L.J.

Fig. 2. A. Ferroportin detection by immunohistochemistry. (A) Ferroportin is found in neurons, oligodendrocytes and astrocytes. The staining of ferroportin is highest in the cortex (A, higher mag- B), thalamus (C), hippocampus (D), cerebellum (E), deep cerebellar nuclei (F) and brain stem (G), (40 Â , scale bar: 500 Am). Antibody to MAP 24 recognized neurons in the cortex (B), hippocampus (D) and cerebellar deep nucleus (F) (arrow, 600x, scale bar: 50 Am ), Purkinje cells in the cerebellum (E, arrow), and oligodendrocytes in the cerebellar white matter (F, asterisk). Peptide competition (MAP 24 peptide) eliminated staining (H). B.

Using fluorescent double staining, ferroportin staining is detected in neurons (A), oligodendrocytes (B), blood vessels (C) and Bergmann glia (D). Red represent the anti-ferroportin Ab, green represents either anti- 111 Neu N, anti-CNPase, or anti-GFAP Ab, respectively. Scale bar: 50 and 20 Am (inset). 112 L.J. Wu et al. / Brain Research 1001 (2004) 108–117 resulting suspension was layered over 5 ml of 1 M sucrose medium L and centrifuged at 140,000 Â g in a Ti 50 rotor in medium L and centrifuged for 30 min at 96,300 Â g in for 1 h to separate any remaining synaptic vesicles from an SW27 rotor. The supernatant was mixed to homogene- soluble proteins. All pellets were then resuspended in 20 ity and centrifuged again for 14 h at 25,000 Â g. The mM Tris–HCl (pH 7.5). The concentrations of total supernatant was collected as synaptic cytosol, and the proteins in each fraction were determined by BCA protein pellets were homogenized in medium L, applied to a assay (PIERCE) against a BSA standard. Fractions were gradient of 7 ml 1.2, 1.0, 0.8, 0.6 and 0.4 M sucrose in separated on 10–20% Tricine-SDS-PAGE, transferred to medium L, and centrifuged for 90 min at 68,000 Â g in an nitrocellulose membrane, and the membrane was sequen- SW27 rotor. Bands at each interface were collected and tially immunoblotted with antibodies against ferroportin washed once with medium L in a Ti50 rotor (45 min at (MAP24, 1:1000), the plasma membrane marker Na+/ 106,500 Â g). The synaptic cytosol was dialyzed against K+ATPase (1:500, Transduction Labs), the synaptic vesicle

Fig. 3. In situ hybridization of ferroportin mRNA. Ferroportin message is found primarily in blood vessels throughout the brain including cerebellum (A), dentate gyrus (B), brainstem (C), striatum (asterisks) and choroid plexus (arrow) (D), and the molecular layer of the hippocampus (E, F, G). Bergmann glia (arrows in A) or choroid plexus (arrow in D) do not show ferroportin mRNA labeling. The dentate granule cells show only faint background labeling (arrow in B). Arrows inC point to transversely cut ferroportin-positive microcapillaries in brain stem. In (E), ferroportin mRNA labeling is shown in a section that was also processed for fluorescent GFAP staining (shown in G). The simultaneous view of ferroportin labeling and GFAP staining (F) demonstrates that astrocytes (white arrows) do not show detectable levels of ferroportin mRNA. Black arrowheads in (E), point to a blood vessel that is ferroportin positive, and that is enveloped by GFAP-positive astrocytic endfeet (white arrowhead). Size bar, A–D 100 Am (shown in B); E–F 50 Am. L.J. Wu et al. / Brain Research 1001 (2004) 108–117 113 marker synaptophysin (1:2000, Chemicon) and the cytosol 3. Results marker h-tubulin (1:2000, Sigma). Diet and iron dextran injection description: Newly 3.1. The distribution of ferroportin in neural cells weaned mice were maintained on a special diet containing either no added iron or normal mouse chow (Fe 0.2 gm/ Antibodies against ferroportin peptides located on the kg chow) from Harlan Teklad for a period of 8 weeks extracellular (MAPs 23 and 25) or intracellular (MAP24) prior to sacrifice. Water was not iron-chelated. We know side of the membrane were used in Western blots of brain from prior experience with these diets that mice on the lysate (Fig. 1A,B). All three antibodies detected a band that low iron diet develop signs of iron deficiency by eight migrated between the 61.3 and 79.6 kD markers. To deter- weeks of age. To further address whether iron manipu- mine the distribution of ferroportin in brain, affinity purified lations could affect brain ferroportin expression, we tried antibodies were used in immunohistochemistry experiments iron-loading mice with several different regimens of iron- on paraformaldehyde-fixed, paraffin-embedded mouse brain dextran injection. These ranged from intraperitoneal injec- sections. Ferroportin is widely expressed in brain, especially tions of 20 mg iron dextran (100 mg/ml solution from in the cortex, hippocampus, thalamus, brain stem and cere- Sigma), daily for 7 days (140 mg total), or for 14 days bellum (Fig. 2). Neuronal cell bodies showed strong positive (280 mg total). Another pair of WT or IRP2 À / À mice staining throughout the brain. The immunoreactivity of was intraperitoneallly (i.p.) injected with 20 mg iron oligodendrocytes was strong in white matter tracts, particu- dextran daily for 3 days (a total of 60 mg iron dextran). larly in the brain stem and cerebellar white matter (F, A second pair was i.p. injected 20 mg daily for 6 days asterisk). In the cerebellum, increased uniform staining was (120 mg). In a final set of experiments, we had two detected in the molecular layer whereas the staining was animals on the iron deficient diet for 9 months, two WT comparatively low in the granular layer (Fig. 2E). Purkinje controls, and two animals that were injected with 2 mg cells expressed ferroportin, mostly in cell bodies and apical iron dextran daily for five successive days. After all of dendrites (Fig. 2E, arrow). Bergmann glia also expressed these dietary manipulations, brain iron was stained with ferroportin. Peptide competition with MAP 24 demonstrated Perls’DAB stain, and iron overload was visibly achieved the specificity of ferroportin detection (Fig. 2H). in all animals injected with iron dextran. Ferroportin The identification of ferroportin in different cell types immunohistochemistry was performed to see if ferroportin was confirmed by double-labeled immunoflurorescence increases or decreases could be discerned at specific sites using antibodies that recognized an oligodendrocyte specific within the CNS on the low iron diet or in iron dextran marker (CNPase), a neuron specific marker (Neu N), or an injected animals. astrocyte specific marker (GFAP) (Fig. 2B).

Fig. 4. Ferroportin protein is found in blood vessels, choroid plexus and ependymal cells. Ferroportin is located on the abluminal side of vessel walls (A) and the foot process of an astrocyte can be seen reaching to the blood vessel (B). Ferroportin is also detected in ependymal cells (C, arrow) and choroid plexus (D), where it is especially concentrated on the ventricular side (v). 114 L.J. Wu et al. / Brain Research 1001 (2004) 108–117

To confirm our immunohistochemistry results, we performed in situ hybridization and detected ferroportin mRNA in cells throughout the brain. In the cerebellum, ferroportin labeling was detected in cells of the granular and molecular layers but not in the Bergmann glial layer (arrows, Fig. 3A). In the dentate gyrus (Fig. 3B), some ferroportin expressing cells were detected, but the granule cells only showed faint background labeling (arrow). The most prominent ferroportin labeling was present in short lines that were identified as microcapillaries (Fig. 3C).In addition, immunoreactive signals were detected in larger size blood vessels throughout the brain, particularly in the thalamus, striatum and cerebellum. Black arrowheads in panel E point to a ferroportin-positive blood vessel that is enveloped by GFAP-positive astrocytic endfeet (white arrowhead). Ferroportin mRNA was not detected in the choroid plexus or ependymal cells (Fig. 3D).

3.2. Ferroportin is expressed in the endothelial cells of the BBB

The primary function of the blood–brain-barrier (BBB) is to maintain a constant microenvironment for brain parenchyma. Transferrin receptors are present on the luminal side of the vascular endothelium to carry iron into the cells [18]. The iron transport mechanism across the abluminal membrane into brain parenchyma is not yet known. To determine whether ferroportin could be a candidate for this role, we examined its distribution within cells. Immunohistochemistry revealed that ferro- portin was located mostly in blood vessel walls (Fig. 4A, B), particularly in the thalamus, striatum and cerebellum. Much of the ferroportin appeared to be concentrated on the abluminal side of vessel walls (Fig. 4B). The process of an astrocyte was immunostained by anti-ferroportin antibody and the foot process was clearly visible extend- ing from the cell body to the blood vessels (Fig. 4B, asterisk). Ferroportin was also detected on the ventricular side of ependymal cells (Fig. 4C) and the choroid plexus (Fig. 4D).

3.3. Ferroportin is present in synaptic vesicles

We noted that ferroportin appeared to be localized in punctate dots that were not associated with cell bodies throughout the cerebellar molecular layer, cerebral cortex, striatum, thalamus and brain stem (Fig. 5A,B). However, ferroportin was not found in white matter tracts such as the corpus callosum and cerebellar white matter. To identify Fig. 5. (A) Ferroportin is present in synaptic vesicles. Ferroportin immuno- the punctate structures in which ferroportin was concen- reactivity is found in punctate spots along the dendrites of Purkinje cells and trated, we performed electron microscopy using pre-em- in the cerebellar molecular layer. (B) The punctate spots are more visible in a bedding immunohistochemistry and we discovered that magnified view. (C–F) Using pre-embedding immunohistocytochemistry ferroportin is associated with presynaptic vesicles (Fig. for electron microscopic studies, ferroportin is detected in presynaptic vesicles (black granules represent areas of ferroportin positivity. Arrows 4C–F). To confirm our ultrastructural identification of point to silver grains on pre-synaptic vesicles in C and D. Single synapses are ferroportin in the presynaptic vesicles, we prepared a shown in (C–E). A larger field containing two synapses (arrows) is shown in subcellular fractionation from crude synaptosomal prepa- (F). Scale bar: 200 nm. L.J. Wu et al. / Brain Research 1001 (2004) 108–117 115 rations. By sucrose density gradient centrifugation, rat cerebral synaptosomes were fractionated into synaptic cytosol, synaptic vesicle, and synaptic plasma membrane fractions and then analyzed by sequential immunoblotting with various antibodies as indicated (Fig. 6). The relative purity of these subcellular fractions was confirmed by immunoreactivity corresponding to markers of synaptic vesicles (synaptophysin), plasma membrane (Na+/K+- ATPase), and cytosol (h-tubulin). Ferroportin was present predominantly in the synaptic vesicle fraction, consistent with our ultrastructural observations.

3.4. The expression level of ferroportin remains unchanged Fig. 7. Expression of ferroportin is not detectably affected by systemic iron by iron overload or depletion status or loss of IRP2 activity. Ferroportin Western blotting was performed on brain lysates from WT or IRP2 À / À animals on a normal or low iron To determine whether ferroportin expression was detect- diet as described in materials and methods. Lysates were also probed with ably regulated by changes in systemic iron status, we actin antibodies as a loading control. To assess CNS iron status and the performed immunohistochemistry, Northern blots and West- effect of dietary manipulations, TfR mRNA levels were evaluated on Northern blots of total RNA prepared from these animals. Although TfR ern blots on animals maintained on a low iron diet after mRNA levels increased in WT animals on the iron-deficient diet, weaning compared to normal diet, and animals injected with ferroportin protein levels did not change. TfR mRNA levels were low iron dextran to produce iron overload (see Materials and and unregulated in IRP2 À / À mice, in accordance with previous observations [21]. W/T: wild type, IRP2 À / À IRP2 knock-out , N: normal iron diet, L: iron deficient diet.

methods). We also evaluated IRP2 À / À mice in which iron status was similarly manipulated, since ferroportin expres- sion is markedly increased in the intestinal mucosa of IRP2 À / À mice [15]. However, we could not detect changes in ferroportin expression, either with immunohis- tochemistry (not shown), or in total brain lysates of WT or IRP2 À / À animals on low or normal iron diets by these measurements (Fig. 7), even though we could detect an increase in TfR mRNA levels in animals on the low iron diet (Fig. 7, bottom panel) indicating that the dietary iron restriction affected the brain. Notably, TfR expression is appropriately increased in the brain of WT animals on a low iron diet [18], but TfR levels are low and unregulated in IRP2 À / À mice, consistent with our previous observations that IRP2 is important for TfR transcript stabilization and regulation in animals [29,30].

4. Discussion

Our studies have revealed that ferroportin is widely expressed throughout the brain. Immunohistochemical anal- ysis showed ferroportin expression by several cell types including neurons, astrocytes, endothelial cells of blood vessels, and epithelial cells of the choroid plexus, whereas Fig. 6. Ferroportin is enriched in the synaptic vesicle fraction of purified ferroportin mRNA was detected by in situ hybridization synaptosomes. One microgram of rat brain synaptosome fractions enriched analysis only in blood vessels. This apparent discrepancy is in synaptic cytosol, synaptic vesicles, and plasma membrane were analyzed likely attributable to the limited detection sensitivity of non- by SDS-PAGE and sequentially immunoblotted with antibodies as isotopic in situ hybridization. However, the in situ hybrid- indicated. Membranes were stripped and reprobed between applications of each antibody. The subcellular localization of ferroportin was determined ization results substantiated the immunohistochemical de- by comparison with markers for synaptic vesicles (synaptophysin), plasma tection of ferroportin in blood vessels, an important site for membrane (Na/K-ATPase), and cytosol (h-tubulin). understanding brain iron homeostasis. 116 L.J. Wu et al. / Brain Research 1001 (2004) 108–117

Immunohistochemistry revealed that ferroportin is However, the presence of ferroportin in synaptic vesicles expressed at regions in which tissues interface with blood implies that axonal free iron can be exported to the synaptic or cerebrospinal fluid, including blood vessel walls, choroid cleft and then be taken up by astrocytes or post-synaptic plexus, and ependymal cells. Iron is generally believed to neurons. These discoveries imply that intracellular iron enter the brain by transport across the blood vessel endo- trafficking within the CNS may represent an important thelial cells that form the blood–brain barrier. After uptake mechanism for moving iron from one location to another. of transferrin-bound iron, iron dissociates from transferrin Moreover, abnormalities in iron homeostasis may have a and enters the brain parenchyma by an unknown mechanism direct adverse effect on synaptic integrity. [18]. Our observation that ferroportin appears more concen- trated on the abluminal side of blood vessels implies that iron can be transported from blood stream to brain tissue. Acknowledgements The expression of ferroportin in astrocytic foot processes further suggests a critical role of ferroportin in brain iron The authors thank the Electron Microscopic Imaging uptake across the BBB. Core Facility of NINDS for their help. Although it is known that choroid plexus epithelial cells contain transferrin receptor, iron, and ferritin, the role of the choroid plexus in regulating iron availability to the CSF has References not been investigated extensively. The localization of ferro- [1] S. Abboud, D.J. Haile, A novel mammalian iron-regulated protein portin on the ventricular side of ependymal cells and the involved in intracellular iron metabolism, J. Biol. Chem. 275 (2000) choroid plexus suggests that ferroportin exports iron from 19906–19912. the brain parenchyma and circulation into cerebro-spinal [2] G.J. Anderson, D.M. Frazer, A.T. McKie, S.J. Wilkins, C.D. Vulpe, fluid, which ultimately reenters the systemic circulation. The expression and regulation of the iron transport molecules Ferroportin mRNA contains an iron-responsive element and IREG1: implications for the control of iron export V from the small intestine, Cell Biochem. Biophys. 36 (2002) 137–146. in its 5 UTR, and the IRE has been previously shown to [3] D. Berg, M. Gerlach, M.B. Youdim, et al., Brain iron pathways and confer regulation of ferroportin expression according to iron their relevance to Parkinson’s disease, J. Neurochem. 79 (2001) status by IRPs [17]. We have previously observed increased 225–236. ferroportin expression in the intestinal mucosa of IRP2 À /À [4] U.V. Berger, M.A. Hediger, Comparative analysis of glutamate trans- mice [21]. However, Western and Northern blots indicate porter expression in rat brain using differential double in situ hybrid- ization, Anat. Embryol. (Berl). 198 (1998) 13–30. that there is no significant difference in ferroportin expres- [5] J.R. Burdo, S.L. Menzies, I.A. Simpson, et al., Distribution of diva- sion in the brain in animals on a low-iron diet, even though lent metal transporter 1 and metal transport protein 1 in the normal an increase in transferrin receptor expression can be shown and Belgrade rat, J. Neurosci. Res. 66 (2001) 1198–1207. under the same conditions. Thus, although regulation of [6] F. Canonne-Hergaux, S. Gruenheid, P. Ponka, P. Gros, Cellular and ferroportin expression in the blood–brain barrier and cho- subcellular localization of the Nramp2 iron transporter in the intesti- nal brush border and regulation by dietary iron, Blood 93 (1999) roid plexus could offer a potential mechanism for regulating 4406–4417. entry of iron into the CNS, we find no evidence that such [7] J.R. Connor, G. Pavlick, D. Karli, S.L. Menzies, C. Palmer, A histo- regulation takes place. It remains possible that ferroportin is chemical study of iron-positive cells in the developing rat brain, regulated at specific sites that are important in brain iron J. Comp. Neurol. 355 (1995) 111–123. homeostasis, but our techniques may be insufficiently sen- [8] D.W. Dickson, W. Lin, W.K. Liu, S.H. Yen, Multiple system atrophy: a sporadic synucleinopathy, Brain Pathol. 9 (1999) 721–732. sitive to detect regulation in a small population of cells. [9] A. Donovan, A. Brownlie, Y. Zhou, et al., Positional cloning of zebra- An unexpected and interesting outcome of our work is fish ferroportin1 identifies a conserved vertebrate iron exporter, Na- that presynaptic vesicles contain ferroportin. This discovery ture 403 (2000) 776–781. implies that iron is transported from cytosol into vesicles in [10] P.R. Dunkley, J.W. Heath, S.M. Harrison, P.E. Jarvie, P.J. Glenfield, the synaptic cleft and that ferrous iron could be released into J.A. Rostas, A rapid Percoll gradient procedure for isolation of syn- aptosomes directly from an S1 fraction: homogeneity and morphology the synaptic cleft upon vesicle fusion. The presence of free of subcellular fractions, Brain Res. 441 (1988) 59–71. iron in the synapse could potentially expose the pre- and [11] N. Gordon, Iron deficiency and the intellect, Brain Develop. 25 post-synaptic membranes to iron-dependent oxidation and (2003) 3–8. damage. The specificity of ferroportin in metal transport has [12] H. Gunshin, B. Mackenzie, U.V. Berger, et al., Cloning and charac- not been experimentally determined. However, specificity terization of a mammalian proton-coupled metal-, Na- ture 388 (1997) 482–488. studies of the related metal importer DMT1 reveal that [13] S.J. Hayflick, Pantothenate kinase-associated neurodegeneration (for- DMT1 transports , cadmium and efficient- merly Hallervorden-Spatz syndrome), J. Neurol. Sci. 207 (2003) ly [12]. Manganese is believed to undergo axonal transport 106–107. and to be concentrated in the synaptic vesicles of glutama- [14] G. Lao, V. Scheuss, C.M. Gerwin, et al., Syntaphilin: a syntaxin-1 tergic cells [26]. It is therefore theoretically possible that the clamp that controls SNARE assembly, Neuron. 25 (2000) 191–201. [15] T. LaVaute, S. Smith, S. Cooperman, et al., Targeted deletion of iron ferroportin detected in synaptic vesicles is there for the regulatory protein 2 causes misregulation of iron metabolism and purpose of transporting manganese. There is no known or neurodegenerative disease in mice, Nat. Genet. 27 (2001) 209–214. postulated role for iron in synaptic vesicles at this time. [16] D. Li, Effects of iron deficiency on iron distribution and gamma- L.J. Wu et al. / Brain Research 1001 (2004) 108–117 117

aminobutyric acid (GABA) metabolism in young rat brain tissues, [25] M. Takanashi, H. Mochizuki, K. Yokomizo, et al., Iron accumulation Hokkaido Igaku Zasshi 73 (1998) 215–225. in the substantia nigra of autosomal recessive juvenile parkinsonism, [17] W.R. Martin, F.Q. Ye, P.S. Allen, Increasing striatal iron content Parkinsonism Relat. Disord. 7 (2001) 311–314. associated with normal aging, Mov. Disord. 13 (1998) 281–286. [26] A. Takeda, Manganese action in brain function, Brain Res. Brain Res. [18] T. Moos, E.H. Morgan, Transferrin and transferrin receptor function Rev. 41 (2003) 79–87. in brain barrier systems, Cell. Mol. Neurobiol. 20 (2000) 77–95. [27] J.P. Tam, Synthetic peptide vaccine design: synthesis and properties [19] C.P. Ponting, Domain homologues of dopamine beta-hydroxylase and of a high-density multiple antigenic peptide system, Proc. Natl. Acad. ferric reductase: roles for iron metabolism in neurodegenerative dis- Sci. 85 (1988) 5409–5413. orders? Hum. Mol. Genet. 10 (2001) 1853–1858. [28] R.B. Wilson, Frataxin and frataxin deficiency in Friedreich’s ataxia, [20] T.A. Rouault, Systemic iron metabolism: a review and implications J. Neurol. Sci. 207 (2003) 103–105. for brain iron metabolism, Pediatr. Neurol. 25 (2001) 130–137. [29] E.G. Meyron-Holtz, M.C. Ghosh, K. Iwai, T. Lavaute, X. Brazzolotto, [21] T. Rouault, R. Klausner, Regulation of iron metabolism in eukaryotes, U. Berger, W. Land, H. Ollivierre-Wilson, A. Grinberg, P. Love, T.A. Curr. Top. Cell. Regul. 35 (1997) 1–19. Rouault, Genetic ablations of iron regulatory proteins 1 and 2 reveal [22] B.D. Schneider, E.A. Leibold, Regulation of mammalian iron homeo- why iron regulatory protein 2 dominates iron homeostasis, EMBO J., stasis, Curr. Opin. Clin. Nutr. Metab. Care 3 (2000) 267–273. in press. [23] Z.H. Sheng, J. Rettig, T. Cook, W.A. Catterall, -dependent [30] S.R. Smith, S. Cooperman, T. Lavaute, N. Tresser, M. Ghosh, E. interaction of N-type calcium channels with the synaptic core com- Meyron-Holtz, W. Land, H. Ollivierre, B. Jortner, R. Switzer, A. plex, Nature 379 (1996) 451–454. Messing, T.A. Rouault, Severity of neurodegeneration correlates with [24] M.A. Smith, P.L. Harris, L.M. Sayre, G. Perry, Iron accumulation in compromise of iron metabolism in mice with iron regulatory protein Alzheimer disease is a source of redox-generated free radicals, Proc. deficiencies, Ann. N.Y. Acad. Sci., in press. Natl. Acad. Sci. U. S. A. 94 (1997) 9866–9868.